m

F Nous contacter


0
     
Post-edited

H 1 Learning with differentiable perturbed optimizers

Auteurs : Berthet, Quentin (Auteur de la Conférence)
CIRM (Editeur )

Loading the player...
supervised learning in ML perturbation methods learning with perturbed optimizers Fenchel-Young losses properties and regularity of the method classification on CIFAR-10 supervised learning to rank supervised shortest paths learning

Résumé : Machine learning pipelines often rely on optimization procedures to make discrete decisions (e.g. sorting, picking closest neighbors, finding shortest paths or optimal matchings). Although these discrete decisions are easily computed in a forward manner, they cannot be used to modify model parameters using first-order optimization techniques because they break the back-propagation of computational graphs. In order to expand the scope of learning problems that can be solved in an end-to-end fashion, we propose a systematic method to transform a block that outputs an optimal discrete decision into a differentiable operation. Our approach relies on stochastic perturbations of these parameters, and can be used readily within existing solvers without the need for ad hoc regularization or smoothing. These perturbed optimizers yield solutions that are differentiable and never locally constant. The amount of smoothness can be tuned via the chosen noise amplitude, whose impact we analyze. The derivatives of these perturbed solvers can be evaluated eciently. We also show how this framework can be connected to a family of losses developed in structured prediction, and describe how these can be used in unsupervised and supervised learning, with theoretical guarantees.
We demonstrate the performance of our approach on several machine learning tasks in experiments on synthetic and real data.

Keywords : perturbation methods; structured learning

Codes MSC :
62F99 - None of the above but in this section
68W20 - randomized algorithms
90C06 - Large-scale problems

Ressources complémentaires :
https://www.cirm-math.fr/RepOrga/2133/Slides/perturbations_berthet.pdf

    Informations sur la Vidéo

    Réalisateur : Hennenfent, Guillaume
    Langue : Anglais
    Date de publication : 06/04/2020
    Date de captation : 09/03/2020
    Collection : Research talks ; Computer Science ; Control Theory and Optimization
    Format : MP4 (.mp4) - HD
    Durée : 00:50:11
    Domaine : Computer Science ; Control Theory & Optimization
    Audience : Chercheurs ; Doctorants , Post - Doctorants
    Download : https://videos.cirm-math.fr/20120-03-09_Berthet.mp4

Informations sur la rencontre

Nom de la rencontre : Optimization for Machine Learning / Optimisation pour l’apprentissage automatique
Organisateurs de la rencontre : Boyer, Claire ; d'Aspremont, Alexandre ; Gramfort, Alexandre ; Salmon, Joseph ; Villar, Soledad
Dates : 09/03/2020 - 13/03/2020
Année de la rencontre : 2020
URL Congrès : https://conferences.cirm-math.fr/2133.html

Citation Data

DOI : 10.24350/CIRM.V.19622903
Cite this video as: Berthet, Quentin (2020). Learning with differentiable perturbed optimizers. CIRM. Audiovisual resource. doi:10.24350/CIRM.V.19622903
URI : http://dx.doi.org/10.24350/CIRM.V.19622903

Voir aussi

Bibliographie

  • PAPANDREOU, George et YUILLE, Alan L. Perturb-and-map random fields: Using discrete optimization to learn and sample from energy models. In : 2011 International Conference on Computer Vision. IEEE, 2011. p. 193-200. - https://doi.org/10.1109/ICCV.2011.6126242

  • KALAI, Adam et VEMPALA, Santosh. Efficient algorithms for online decision problems. In : Learning Theory and Kernel Machines. Springer, Berlin, Heidelberg, 2003. p. 26-40. - http://dx.doi.org/10.1007/978-3-540-45167-9_4

  • BERTHET, Quentin, BLONDEL, Mathieu, TEBOUL, Olivier, et al. Learning with Differentiable Perturbed Optimizers. arXiv preprint arXiv:2002.08676, 2020. - https://arxiv.org/abs/2002.08676



Z